> From: "Christian Huitema" <huit...@huitema.net> > Indeed, any substantial change in routing and addressing would most > probably end up changing all hosts, or at least most of them.
Not necessarily, provided that the identity/location mapping layer is deployed as a set of middleboxes (as opposed to being instantiated in all the hosts). One of the benefits from that deployment model is that a new location namespace can be deployed invisibly to the hosts. And of course the routing is invisible to the hosts at the moment anyway, so changing that without the hosts noticing is almost inevitable. > I also think that a research group should not necessarily shy away from > investigations that affect all hosts. If research comes out with a > brilliant shiny star on the horizon, then engineering might follow. You're correct about that. I guess I didn't clearly distinguish in my previous message between research and engineering. From the engineering perspective, a design which requires changes to all hosts is, IMO, a non-starter. But at the same time, I definitely am already a follower of the model that, if all we do is "hack on the edges of current solutions" without some larger vision of where we are going in the long term (and that's often, or usually, missing, in many of the incremental changes), those many small incremental changes will result in nothing more than 'architecture cancer'. So, to me, a "brilliant shiny star on the horizon" really is a requirement; you have to know where you're going in the long term, and that destination has to be a clear, clean, functional architecture. So perhaps we're not so far apart, actually? But in the end, as a practical matter, to get anything done, the path there has to consist of many small steps, each of incremental nature... Noel _______________________________________________ rrg mailing list rrg@irtf.org http://www.irtf.org/mailman/listinfo/rrg